Local floras are a basis of biogeography, ecology, evolution, and systematics, and they add value to research sites. Mountain Lake Biological Station is located between 1,150 and 1,319 m elevation in southwestern Virginia on infertile, acidic soils supporting a second-growth forest strongly dominated by Quercus rubra about 150 years old. We sampled vascular vegetation on 352 randomly distributed plots 10m in diameter with 8 subplots of 1 m² each. The plots contained 175 taxa (including 7 taxa that included 16 lumped species); other fieldwork added 43 species for a total of 227 species in the forest. We excluded disturbed areas around buildings, roadsides, and ponds that harbor many ruderal and invasive species. Average detection probability of a species on a plot was estimated to be 75%. We recorded binary occurrence (present or absent on a plot) and prevalence (occurrence on 0, 1, 2, … 8 subplots per plot) and a set of environmental variables measured in the field or laboratory, or estimated from LIDAR data collected by NEON. Nonmetric multidimensional scaling ordination rejected the null hypothesis of random distributions of herbaceous species among plots. Cluster analysis defined eight groups of plots based on prevalence of herbaceous species to illuminate rejection of the null hypothesis. The environments of these clusters, determined only by their floras, demonstrated the importance of rock type, soils, solar radiation, canopy height, and topography in determining the local distribution of species. Significant differences among assemblages and their environments occurred within a superficially homogeneous forest.